00 Basic plan to new signups. Remaining Basic subscribers were auto-migrated to pay-per-use starting June 1, 2026. Pro at $5,000 per month and custom Enterprise contracts remain for high-volume and streaming use cases. For everyone else, the X API pay-per-use model is the baseline, which is a win for small projects but introduces new cost traps for production apps.
There is no published hard cap on writes. The practical ceiling is whatever monthly spending limit a developer sets in the console. Reads are a different story. A global ceiling of 2 million post reads per month applies on pay-per-use. That limit protects infrastructure and pushes sustained high-volume reading to Enterprise.
For teams updating internal budgets and compliance sheets, treat this as the definitive Twitter API pricing 2026 snapshot: new developers pay only for what they call, not a flat subscription. The shift changes how to estimate costs. Instead of counting projects by endpoints allowed, plan by resources returned and how link-heavy the posting workload is.
Pay-per-use charges depend on the resource and what is returned. Reads are billed per resource returned, not per request. For example, a search that returns 100 posts is charged for 100 post reads, not one API call. The platform applies a monthly ceiling of 2 million post reads on pay-per-use. Writes do not have a published hard cap, so budget limits are the control. XMCP uses the same meter and requires a developer account. In short, MCP server billing follows the same pay-per-use rules.
Credits are prepaid. Developers load X API credits into an account, each call debits that balance, and calls fail at zero. Monthly spending limits can be set per project and should be enabled on day one. X offers a rebate program that returns up to 20 percent in xAI compute credits on qualifying spend above
,000. Those rebates are not X API credits, so they do not reduce future X API bills.
Monthly cost scenarios
Real budgets depend on workload shape. Three quick scenarios help set expectations. A lightweight brand account posting 10 plain-text tweets per day for a month spends about $4.50, since those are all standard writes. A link-heavy campaign that publishes 300 posts with URLs comes to about $60 in writes alone because link posts are charged at $0.20, roughly 13 times the plain-text rate. An AI agent that summarizes trends and posts updates can end up near $65.70 per month, with most spend coming from reads, not writes.
Key cost drivers to watch:
- Links in posts charge the 13x write fee at $0.20 per post.
- Reads compound quickly because billing is per resource returned, not per request.
- Search queries, timelines, and multi-step enrichment inflate read totals.
- Retries, queues, and A/B tests multiply writes if not deduped.
Stop doing per-tweet math
Per-post arithmetic hides the real spend. Think in workflows: data acquisition, enrichment, decisioning, and publishing. Calculate the average resources returned per search and timeline pull, then layer in write retries, queues, and scheduling. Preload X API credits with a margin, set a monthly cap, and export usage logs weekly. That approach turns ad hoc guessing into predictable spending.
Reduce spend or go flat-fee
Why is the X API expensive now? Two levers drive it: fraud and link abuse controls, and the value of the underlying social graph. The 13x fee on URL posts discourages spam and funds link scanning and safety review. Read ceilings reduce scraping and nudge sustained high-volume consumers toward contracts that include compliance and dedicated support.
Optimizing for pay-per-use:
- Filter aggressively on search so each call returns fewer, higher-quality posts.
- Cache user objects and post metadata to avoid repeated lookups.
- Prefer native content and threads when possible, reducing paid link posts.
- Add backoff, deduping, and idempotent keys to eliminate duplicate writes.
- Run scheduled data pulls during off-peak automation windows and batch pagination.
- Set per-project caps, alert at 50, 75, and 90 percent of monthly budgets.
Flat-fee posting is the alternative. A relay service posts on a developer's behalf under a fixed monthly plan, with no OAuth dance and no per-call math for posting. Typical tiers run around
1.99 per month for a basic plan and 9 per month for advanced features, with a short free trial. Flat-fee is advantageous for
link-heavy publishing, scheduling with retries and queues, and high-variance workloads. For plain-text, low-volume posting, pay-per-use generally costs less.
Related tools and guides:
- X Developer Console: Billing and usage exports for precise tracking.
- Official rate cards: Current per-endpoint definitions and read limits.
- Budget worksheets: A simple spreadsheet that multiplies expected resources per workflow by unit rates.
- Growth without API code: For founders and creators who prefer audience results over API math, BeeReach (https://beereach.io) provides safe, human-like engagement on X with precise Radar Filtering, a full client portal, Content Scheduler, and Smart Automation Rules. Plans start at $39 per month.
FAQ
Is the Twitter API free in 2026?
No. The free tier for new developers ended on February 6, 2026. Legacy Basic accounts were auto-migrated to pay-per-use starting June 1, 2026, with Pro and Enterprise reserved for high-volume needs.
How do X API credits and spending limits work?
Credits are prepaid. Every request debits the balance by resource returned or write type, and calls stop when the balance hits zero. Set a monthly spending limit on day one so write bursts and unexpected reads cannot exceed budget.
When is flat-fee better than pay-per-use?
Flat-fee posting wins when workloads are link-heavy or need scheduling with retries and queues, since link writes cost $0.20 each under pay-per-use. If posting is mostly plain text at low volume, the pay-per-use model usually stays cheaper while preserving full control over reads and enrichment.